Secondary structure prediction. Amino acid sequence -> Secondary structure Alpha helix Beta strand Disordered/coil 70% accuracy 1991, 81% accuracy in

Secondary structure prediction


Amino acid sequence -> Secondary structure

Alpha helixBeta strandDisordered/coil

70% accuracy 1991, 81% accuracy in 2009


Limits:

Limited to globular proteins

Not for membrane proteins


Applications

Site directed mutagenesisLocate functionally important residuesFind structural units / domains


Techniques

Linear statisticsPhysicochemical propertiesLinear discriminationMachine learningNeural NetworksK-nearest neighboursEvolutionary treesResidue substitution matrices

Using evolutionary information = Multiple sequence alignments.

Secondary structure predictionJnet

Cuff and Barton (2000)

Neural Network

Training set: 480 proteins (non homologous)

Construction of MSA for each using BLAST


Neural network

Ni Neuron i, Nj neuron jWij weight from Ni to Nj

Ni NjWij

Signal forward propagationOutput from Ni * Weight Ni to NjInput to Nj is Ij = Oi * Wij


Neural network

Ni Neuron i, Nj neuron jWij weight from Ni to Nj

Ni

Nz

Wij

Nj

Nk

Input layer

Output layer


Neural network

The network receives input values

Ni

Nz

Wiz

Nj

Nk

Wjz

Wkz

Ii

Ij

Ik

Input layer

Output layer


Neural network

Signal forward propagation

Ni

Nz

Wiz

Nj

Nk

Sum of outputs from Ni, Nj, Nk

Wjz

Wkz

Oz

Ii

Ij

Ik


Neural network

Compute the error

Ni

Nz

Wiz

Nj

Nk

Desired value is 1, Oz is 0.8Error is = Oz – desired value = 1-0.8 =0.2

Wjz

Wkz

Oz

Ii

Ij

Ik


Neural network

Error backpropagation

Ni

Nz

Wiz

Nj

Nk

Weights are modified so that the result is a bit closer to what we wanted

Wjz

Wkz

Oz


Neural network

Ni

NzNj

Nk

Input layer

Output layer

Hidden layer

Np

Nq


Neural network

Input layer: read a sequence

A

C

D

…Y

CTEIL...


Neural network


A

C

D

…Y

CDEKL...CDEKL...

010

0


Neural network


A

C

D

…Y

CDEKL...CDEKL...

001

0


Neural network


A

C

D

…Y

CDEKL...CDEKL...


Neural network

Output layer: structureDesired output: known structure

A

C

D

…Y

CDEKL...

b

c

a

Alpha-helix

1

0

0


Neural network

Output layer: structureDesired output: known structure

A

C

D

…Y

CDEKL...

b

c

a

Alpha-helix

1

0

0

0.4

0.6

0.1


Neural network

Error backpropagation = weights are modified

A

C

D

…Y

CDEKL...

b

c

a

Alpha-helix

1

0

0

0.4

0.6

0.1


Jnet architecture

LAPEDCDEKLKLEPNAC

b

c

a

Sequence to structure network

Input layer = window of 17 residuesHidden layer = 9 neuronsOutput layer = 3 neurons


Jnet architecture

FLAPEDCDEKLKLEPNACW

b

c

a

ccaacaaccbbbbbcbbbc


Jnet architecture

FLAPEDCDEKLKLEPNACW

b

c

a

Structure to structure network

b

c

a

Input layer = window of 19 residuesHidden layer = 9 neuronsOutput layer = 3 neurons

ccaacaaccbbbbbcbbbc

ccaaaaacccbbbbbbbbc


Cole et al (2008) Nucleic Acids Research

Geoff Barton, University of Dundee


Uses algorithm Jnet2.0Three state predictionAlpha, beta, coilAccuracy 81.5% (2008)But if no homolog (orphan sequence) 65.9%!

PSIBLAST PSSM matrixHMMer profiles (instead of aa frequencies)Multiple neural networks100 hidden layer units

Jpred


First, search against PDB sequences using BLAST (but only for warning)

PSIBLAST search of UniRef90, 3 iterations, Alignment of hits (filtered at 75% id)

Profiles from alignment (PSSM and HMMer)

Profiles are input to JNet

Alternative: user provides alignment (faster)

Jpred

Secondary structure predictionAdvanced Jpred4 usage


Jpred output


Jpred output / JalviewJPred


Jpred output / view allJPred


Jpred output / PDF outputJPred

Starting Jalview

Open Firefox with JRE (from ZDV)

Go to http://www.jalview.org

Click the pink arrow “Launch Jalview Desktop”

You can close all the demo windows that appear

Exercise 1/4Jalview 2D prediction

http://www.jalview.org/

Load an alignment

Use MR1_fasta.txtThis is an alignment of a fragment of the mineralocorticoid receptor

Open it from File > Input alignment > From file(Hint: You can load it directly as an URL, e.g. https://cbdm.uni-mainz.de/files/2015/02/MR1_fasta.txt)

The alignment has its own Menu tabsTry Colour > Clustalxto see conservation


https://cbdm.uni-mainz.de/files/2015/02/MR1_fasta.txt

https://cbdm.uni-mainz.de/files/2015/02/MR1_fasta.txt


Build a phylogenetic tree

Also using MR1_fasta.txt

Try Calculate > Calculate tree > Neighbour joining using PAM 250

Hint: We used identifiers indicating species names

Try now Calculate > Calculate tree > Average distance using PAM 250

Compare results. Which one is better?


XenopusAnolis

Taeniopygia

GallusMonodelphis Danio

Web service -> Secondary Structure Prediction -> Jnet secondary str pred

No selection (or all sequences selected) = Jnet runs on top sequence using the alignment (fast)

One sequence (or region) selected = Jnet runs on that sequence using homologs (slow)

Some sequences selected = Jnet runs on top one using homologs (slow) Try with all sequences selected


If this doesn’t work you can run directly MR1_fasta.txt on jpred4.

Use the advanced optionUpload a file optionSelect type of input = Multiple alignment (use format FASTA)Tick the skip PDB search optionThere is an option to view output in Jalview


Annotations: • Lupas_21, Lupas_14, Lupas_28

Coiled-coil predictions for the sequence. 21, 14 and 28 are windows used.


Annotations: • JNETHMM, JNETALIGN: predictions using diff profiles• Jnetpred: Consensus prediction. Beta sheets: green arrows. Alpha helices: red tubes.


Annotations: • JNETCONFConfidence in the prediction.


Annotations: • JNETSOL25,JNETSOL5,JNETSOL0

Solvent accessibility predictions - binary predictions of 25%, 5% or 0% solvent accessibility.


Obtain the sequence of the human glutamine synthetase.

Run BLAST with the human sequence against: 1) the archaea Methanosarcina2) the bacteria Escherichia coli3) the fungi Pseudozima Antarctica

Get the best homolog, align the sequences (including the human protein, on top) and use the input in Jalview.

Exercise 4/42D prediction of known 3D

NCBI BLAST against single species is faster!


Obtain the sequence of the human glutamine synthetase.

Run BLAST with the human sequence against: 1) the archaea Methanosarcina2) the bacteria Escherichia coli3) the fungi Pseudozima Antarctica

Get the best homolog from each, align the sequences and use the input in Jalview. Put the human protein on top.


Load the alignment in Jalview and run web prediction.

Alternative. Run the alignment in the Jpred4 server. (Hint: You could run the human sequence alone but that will search for homologs and will take very long)

Compare the prediction with the known 3D of the human protein (open it in Chimera, File > Fetch by ID > PDB 2QC8)


We need to hide all chains except one.

Select one of the chains (ctrl + click on a residue, then arrow up). Invert selection (press arrow right).

Actions > Ribbon > HideActions > Atoms/bonds > Hide

Select the chain and focus on itActions > Focus


Compare the output of jpred/jalview 2D pref with the 3D structure of this protein.

For example, locate a predicted helix or beta-strand in Jalview. Find out the start and end positions hovering over the human sequence with the mouse (the numbers on top of the alignment are different from the amino acid positions in each sequence).

Color the corresponding residues it in the 3D view usingSelect > Atom specifierAnd ranges: e.g. :113-126 (predicted as helix)Actions > color > red

Apply color some helices red and strands in green.

Do you see differences? Where are they?Would you say that the 2D prediction was reasonable?


Documents

Secondary structure prediction. Amino acid sequence -> Secondary structure Alpha helix Beta strand Disordered/coil 70% accuracy 1991, 81% accuracy in